context of an advanced organic agriculture of the municipality of Sentmenat (Catalonia, Spain) that circa 1860 had adopted a partial vineyard specialization.
As we said before, there is a vast amount of socio-metabolic research for this case study (Cussó et al., 2006; Garrabou et al., 2010, 2008, Marull et al., 2010; Olarieta et al., 2008; Tello et al., 2016, 2012), based on an array of documentary sources such as the Estudio Agrícola del Vallès (Garrabou and Planas, 1998).
In this chapter, we will only focus to Sentmenat, because we have data on the Population Census, Cadastral Map, and Land-Use Register (Amillaramiento), which allow us to infer the characteristics of the Domestic Units.
Sentmenat is located along the tectonic boundary between the Vallès plain and the Catalan pre-littoral mountain range.
It comprises 2,880 ha with a great topological, geological and soil diversity affected by a sub-humid Mediterranean littoral climate. It is located in the middle of our broader case study of the Vallès (Figure 2.3).
Figure 6.2 shows how c.1860 vines coexisted with dryland polycultures and woods forming agricultural, pasture and forest mosaics. Population density was 60 inhab./km2 in the mid-nineteenth century, which according to Badia-Miró and Tello (2014) fits demographic optimal conditions for vineyard specialization. Vineyards covered 42% of the total surface, 72% of cropland, and produced some 17,000 hl of wine a year. Although no data on vineyard surface is available for 1890 at the time when it reached its peak, we know that wine production increased up to 26,000 hl (Planas, 2015) during the period when the Phylloxera Plague started destroying French vines in the 1860s and ended up devastating those of the Valles in the late 1880s.
Afterwards, very few vineyards were recovered (Badia-Miró et al., 2010).
Most important, until 1860 the agroecosystem functioning of Sentmenat was still mainly restricted at regional level from the reproductive view of its funds’ maintenance. Local food and firewood needs were still largely met at regional level, while livestock and soil fertility maintenance were kept strongly integrated one another at local level (Chapter 4).
Figure 6.2. Land-uses in Sentmenat circa 1860. Source:
Our own.
Chapter 6. Beyond Chayanov, SFRA c.1860
4. Methodology
4.1 The Sustainable Farm Reproduction Model for peasant units
In order to obtain different profiles of possible biophysical flow distributions in traditional organic farm systems, we develop a socioecological model based on the ability to reproduce their basic funds under the prevailing natural and technical endowments (Figure 6.1).
We adopt the farmers’ standpoint, and set the scope and temporal boundaries of the case study in the municipality of Sentmenat along a year c.1860, following the analytical frame proposed by Tello et al. (2015, 2016) based on a fund-flow approach (Mayumi, 1991), and the ensuing graph modelling put forward in Chapter 3. The funds considered are, as mentioned: domestic unit (DU), domesticated species (livestock), and farmland (soil). Through a socio-metabolic linear programming we account for the multiple interactions (flows) set among these three funds (Figure 6.3).
The interest of this SFRA model lies on how farmland can be optimized through the allocation of possible land-uses. To establish which size one fund should be within a farm, we consider boundary conditions the size of the other two funds, despite having to double check if in the optimized scenario livestock density changes would increase efficiency on resource allocation. This is because through linear programming we are constrained and for calculating the optimum dimension of one fund, both the others have to be fixed.
Therefore, we start the accountancy by defining the size and composition of the DU, and this involves an important decision that affects interpretation. The representative family selected supposes a rate of consumers/producers consistent with the average dependence rate of the whole municipality, so as to allow us to extrapolate the DU results at local level. Indeed, following Chayanov ([1925]1986), we also run five different stages of the life cycle of this representative family in order to see how requirements would change over time, and not under or overestimate the capacity of land to host population. Thus, we will only present the results for the DU composition stage that reaches highest requirements.
Figure 6.3. Modelling diagram for the SFRA c.1860. Source: our own. Squares represent funds and arrows the flows interlinking them. Grey arrows are fluxes constrained as boundary conditions, while black ones are the restrictions
calculated with the optimization model. Discontinuous lines emphasize the objective functions to optimize.
Chapter 6. Beyond Chayanov, SFRA c.1860
4.2. The linear programming model28
The mathematical procedure for running the model is linear programming. This method achieves the best result by maximizing or minimizing (optimize) a linear first-degree function, allowing infinite variables and constraints as long as they are linear. Once the model is defined, it is run through Simplex algorithm programmed using Java software Gusek. The model comprises main variables (22), secondary variables (128), parameters (3), constraints (105) and objective functions (3). In this initial case study, we cannot implement a sensitivity analysis for optimization due to the lack of data owing to the historical character of the case study. Therefore, results will have to be considered cautiously, but we think that still have explanatory effect. We consider it to be relevant for further studies to guarantee more consistent results.
4.2.1. Variables and parameters
The main variables, from X1 to X22, represent the surface of each land-use.
Each secondary variable (Xi,j) belongs to a use Xi, and represents a fraction of the total surface of land-use. As pointed in section 4.1, dimensions of DU and livestock heads are boundary conditions. We then define a parameter Z that represents the total number of members in the farm unit. When results for the whole municipality are extrapolated, the surface will be accounted by weighting farmland surfaces of each Z for its frequency in census and cadastral data. Small livestock numbers depend on the size of the DU (Z), while draught animals depend on the cropping surface through another parameter M representing its density.
4.2.2. Constraints
All variables are subject to a number of biophysical constraints expressed by linear inequalities. These restrictions contain boundaries imposed by the conditions for the agroecosystem reproduction, ranging from obtaining enough food, fuel and money, feeding livestock, closing nutrient soil cycles, to keeping the historically prevailing crop rotations. We define minimum and maximum flows for the three funds, which allow each of them to ensure agroecological sustainability (minimum for inputs, and maximum for outputs of the fund). Table 6.1 summarizes the main aspects considered in the programming model, from which constraints are derived.
Regarding DU consumption, we consider basic human material needs as determined ex-ante, and not considering the improving of farmers’ well-being (Harrison, 1975). Given that we are setting a threshold for farming family reproduction, and not on how to organize farm counting with abundant means of production, we consider this demand to be inelastic. Indeed, we estimated the ability of a DU for doing labour at monthly scale, versus requirements from farmland and livestock. And for consumption we estimated the requirements of the typical diet (Cussó and Garrabou, 2012), the fuel for heating (Marco et al., 2017) and monetary requirements for clothing, footwear, housing, tools and tax burdens (Colomé, 2015, 1996; Vicedo et al., 2002).
About livestock, we consider the energy requirements for its maintenance and work, mainly from Church (1984), as well as their stall bedding for the barns (Soroa, 1953). However, we do not include here any specific diet but we only estimate the whole sources of feed and left the model decide them according their requirements. As well, an important flow from livestock is manure, which will be fundamental for soil nutrient restoring.
28 We explain the whole model with their variables, assumptions, boundary conditions and constraints, in Annex II.
Chapter 6. Beyond Chayanov, SFRA c.1860
Table 6.1. Constraints and main sources considered in the programming model. Source: our own.
Domestic Unit Livestock Farmland-soil
Food (Cussó and Garrabou, 2012) Feed (Church, 1984; Roca, 2007)
Nutrient balances (González de Molina et al., 2010) Fuel (Marco et al., 2017) Stall bedding (Soroa, 1953) Soil qualities
Money (Colomé, 2015; Vicedo et al., 2002) Irrigated land
Labour (Marco et al., forthcoming)
Finally, regarding soil fund, we assess three dimensions: nutrient balances, total available surface, distribution of soil qualities and yields and the total amount of irrigated land. Nutrient balances involve many different flows (both natural and anthropic), and sources of nutrients. As well, we want to note that here we only consider nutrient balances for cropped areas. Instead, for forest and pastures we continue with the proposal stated in chapter 2 of ensuring a flow of biomass lower that estimated net biomass production within a year.
For land distribution we consider, in order to extrapolate results to the whole municipality, farm surface have to be representative of the constraints with regard to soil qualities.
Behind this guarantee on having the same surface of soil qualities as in the total surface, there is a strong assumption that we made in order to be able to carry this modelisation. That is soil quality valuations for different crops are interchangeable, i.e. that a vineyard in 1st soil quality can change to cereal with the yields associated to 1st soil quality. This assumption obviously strongly affects the results, mainly in the case of changes from extensive to more intensive uses. Therefore, we will be cautious when inferring some conclusions on land-use changes, assessing at least what would it have supposed.
3.2.3. Objective functions
We optimize the resulting model according to the goals farmers might have adopted under the conditions set in each agroecosystem context. Therefore, we consider three functions that characterize different farming objectives: i) the minimum surface to ensure a reproducible exploitation, called ‘intensive optimum’ (Eq. 1); ii) the area required for the reproduction minimizing total labour, or ‘extensive optimum’ (Eq. 2); and iii) the maximum sustainable wine-growing area, or ‘monetary optimum’ (Eq. 3).
h=?W∑3737 i3Z (Eq.2)
h=?W∑3737 j3i3+ C9i3, d, k;Z (Eq.3)
hEl9i%; (Eq.4)
Where Xi is the area of each land-use, Wi the required workdays for each land-use, f(Xi,M,Z) the workdays associated with fertilization practices resulting from the model, and X19
the vineyard area.
Chapter 6. Beyond Chayanov, SFRA c.1860
3.3. Three different farm goals and scenarios
As we have seen in the previous section, we have run the model for the three different scenarios, so as to then compare them with the historical profile. The first function corresponds to the Minimum Reproduction Unit (MRU), which sets the minimum surface and land-use composition a sustainable farm should have. The profile is first set by seeking the area required to meet the needs of each fund (DU, draught animals, animals for consumption, soil fertility), and then by identifying both LACAS (Guzmán et al., 2011) and the landscape synergy resulting from the integration among funds (Lemaire and Franzluebbers, 2014; Marull and Tello, 2010).
The second objective function determines the Peasant Reproduction Unit (PRU), or the amount of surface and land-use composition that minimizes the labour required in the farm. This criterion minimizes the labour required to obtain the same product as MRU (Tannenbaum, 1984).
Then, for both PRU and MRU we study the energy efficiency of the farm by calculating the Final Energy Return on Investment and Final Energy Return on Labour obtained (Tello et al., 2015, 2016). Both profiles allow us to define a range of possible sustainable eco-functional intensification, under a theoretical condition of equal access to land and livestock. Then we use these counterfactual scenarios to identify the potential institutional forces that prevented the achievement of those agroecological optimums.
The third farm scenario is the Maximum Sustainable Specialization (MSS). This would have been the amount of vineyard land that might have been sustained under a reproduction condition, maintaining population density. Hence, we can figure out what the possible evolution of vineyard specialization would had been in comparison to the real one, revealing its biophysical limits, and helping us to identify the social driving forces at stake.
5. Results and Discussion: Drivers of agricultural change in each scenario